colorize_reconstruction produces a reconstruction file same size, runs just few seconds

[alkavan]

So here's the pipeline for my reconstructions:

create_calibration_file_from_exif \
-images=img/* \
-output_calibration_file=calibration

build_reconstruction \
-reconstruction_estimator=GLOBAL \
-calibration_file=calibration \
-descriptor=SIFT \
-feature_density=DENSE \
-images=img/* \
-matching_working_directory=matches/ \
-output_reconstruction=model \
-reconstruct_largest_connected_component=false \
-shared_calibration=true
-num_threads=4

colorize_reconstruction \
-image_directory=img/ \
-input_reconstruction_file=model-0 \
-num_threads=4 \
-output_reconstruction_file=model_color

So it seems that colorize_reconstruction produced the same reconstruction file... what does it mean? it also takes just a few seconds to run, and I'm assuming it does not actually do anything related to color processing? Am I confusing or missing something here? And on the same subject, it would be nice to have a basic README.md in the main repository with some basic pipeline examples, I only figured out what to do because I used OpenMVG and the -help flag ...

[sweeneychris]

As you can see from the command, you've provided an --output_reconstruction flag where the colored reconstruction will write to. I don't know why you assume that an app called colorize_reconstruction does nothing related to color processing... obviously it does.

There is a ton of documentation online www.theia-sfm.org, every app is documented inside the code, and (as documented) you can run any binary with --helpshort to get the command line options. Please consider doing a bit of investigation into the existing resources before claiming the apps do nothing or more documentation is needed. I put a lot of work into documenting the code and the website so please utilize those resources.

On Tue, Nov 13, 2018 at 11:34 PM Igal Alkon [email protected] wrote:

So here's the pipeline for my reconstructions:

create_calibration_file_from_exif
-images=img/*
-output_calibration_file=calibration

build_reconstruction
-reconstruction_estimator=GLOBAL
-calibration_file=calibration
-descriptor=SIFT
-feature_density=DENSE
-images=img/*
-matching_working_directory=matches/
-output_reconstruction=model
-reconstruct_largest_connected_component=false
-shared_calibration=true -num_threads=4

colorize_reconstruction
-image_directory=img/
-input_reconstruction_file=model-0
-num_threads=4
-output_reconstruction_file=model_color

So it seems that colorize_reconstruction produced the same reconstruction file... what does it mean? it also takes just a few seconds to run, and I'm assuming it does not actually do anything related to color processing? Am I confusing or missing something here? And on the same subject, it would be nice to have a basic README.md in the main repository with some basic pipeline examples, I only figured out what to do because I used OpenMVG and the -help flag ...

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sweeneychris/TheiaSfM/issues/221, or mute the thread https://github.com/notifications/unsubscribe-auth/AAwytnLpAIGtmqUqtnsplvNPRJHzC3SSks5uu8eEgaJpZM4YdLTQ .

[charlescva]

Agreed. The documentation is very good as is the code base. I can concur that be colorization code absolutely does work in the 0.8 release.

On Wed, Nov 21, 2018, 2:13 PM Chris Sweeney <[email protected] wrote:

As you can see from the command, you've provided an --output_reconstruction flag where the colored reconstruction will write to. I don't know why you assume that an app called colorize_reconstruction does nothing related to color processing... obviously it does.

There is a ton of documentation online www.theia-sfm.org, every app is documented inside the code, and (as documented) you can run any binary with --helpshort to get the command line options. Please consider doing a bit of investigation into the existing resources before claiming the apps do nothing or more documentation is needed. I put a lot of work into documenting the code and the website so please utilize those resources.

On Tue, Nov 13, 2018 at 11:34 PM Igal Alkon [email protected] wrote:

So here's the pipeline for my reconstructions:

create_calibration_file_from_exif
-images=img/*
-output_calibration_file=calibration

build_reconstruction
-reconstruction_estimator=GLOBAL
-calibration_file=calibration
-descriptor=SIFT
-feature_density=DENSE
-images=img/*
-matching_working_directory=matches/
-output_reconstruction=model
-reconstruct_largest_connected_component=false
-shared_calibration=true -num_threads=4

colorize_reconstruction
-image_directory=img/
-input_reconstruction_file=model-0
-num_threads=4
-output_reconstruction_file=model_color

So it seems that colorize_reconstruction produced the same reconstruction file... what does it mean? it also takes just a few seconds to run, and I'm assuming it does not actually do anything related to color processing? Am I confusing or missing something here? And on the same subject, it would be nice to have a basic README.md in the main repository with some basic pipeline examples, I only figured out what to do because I used OpenMVG and the -help flag ...

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sweeneychris/TheiaSfM/issues/221, or mute the thread < https://github.com/notifications/unsubscribe-auth/AAwytnLpAIGtmqUqtnsplvNPRJHzC3SSks5uu8eEgaJpZM4YdLTQ

.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sweeneychris/TheiaSfM/issues/221#issuecomment-440779462, or mute the thread https://github.com/notifications/unsubscribe-auth/AFGI5jPh-QuI_5a2A9oLmvecWD6FtDJqks5uxaXvgaJpZM4YdLTQ .